Collaborative filtering refers to a class of tools and mechanisms that allow the retrieval of predictive information regarding the interests of a given set of users starting from a large and yet undifferentiated mass of knowledge. Collaborative filtering is widely used in the context of recommendation systems. A well-known category of collaborative algorithms is matrix factorization.
The fundamental assumption behind the concept of collaborative filtering is that every single user who has shown a certain set of preferences will continue to show them in the future. A popular example of collaborative filtering can be a system of suggested movies starting from a set of basic knowledge of the tastes and preferences of a given user. It should be noted that although this information is referring to a single user, they derive this from the...